Multimedia Tools and Applications最新文献

Nowadays, the pipeline system has the safest, most economical, and most efficient means of transporting petroleum products and other chemical fluids. But, the faults in pipelines cause resource wastage and environmental pollution. Most of the existing works focused either on the surface Crack Detection (CD) or Leakage Detection (LD) of pipes with limited features. Hence, efficient crack detection and leakage monitoring are proposed based on the Acoustic Emission (AE) signal and AE image features using a new Berout Retina Net (BRetN) and Tent Chaotic Kaiming-centric Long Short Term Memory (TCK-LSTM) methodologies. The process initiates from the gathering of input data, followed by preprocessing. Then, the cracks are detected by utilizing Berout Retina Net (BRetN), and the features of AE signals are retrieved. On the other hand, the AE signal is transformed into an AE image using Continuous Wavelet Transform (CWT). Further, the AE image features are extracted, followed by the integration of both the AE signal and AE image features. Further, the optimal features are chosen by using Gorilla Troops Optimizer (GTO). Eventually, the TCK-LSTM model is used for detecting the leakage level of the pipeline. The experimental outcomes illustrated that the proposed framework detected crack and leakage levels with 98.14% accuracy, 95.37% precision, and 98.84% specificity when analogizing over the existing techniques.

Detecting deception in videos has been a challenging task, especially in real world situations. In this study, we extracted the facial action units from the micro-expression, and then calculated the frequency and the number of occurrences of each action unit. To get more information on different scales, we proposed a combination scheme of Multi-Scale Feature (MSF) model and Multi-Head Attention (MHA). The MSF model consists of two CNN with different convolution kernels and GELU is used as the active function. The MHA model was designed to divide the input features into different subspaces and generate attention for each subspace to make the features more effective. We evaluated our proposed method on the Real-life Trial dataset and achieved an accuracy of 87.81%. The results show that the MSF and MHA model could increase the accuracy of deception detection task. And the comparative experiment demonstrates the effectiveness of our proposed method.

Adaptive learning is an educational methodology that allows the personalization of learning according to the student’s pedagogical path. In digital environments, the strategic use of technologies enhances adaptive learning initiatives, enabling a dynamic understanding of intricate contextual nuances and the ability to identify and recommend appropriate learning activities. Therefore, this work proposes developing and evaluating a prototype that uses a large language model to create adaptive educational activities in face-to-face and virtual environments automatically. The applied methodology involves the implementation of a large language model with advanced cognitive capabilities to generate learning activities that adapt to individual needs. A proof of concept was developed to evaluate the practicality and usability of this approach. The research results indicate that the approach is practical and adaptable to different educational contexts, reinforcing the synergy between adaptive learning, artificial intelligence, and learning environments. The proof of concept evaluation showed that the prototype is highly usable, validating the proposal as an innovative solution to the growing needs of modern education.

Identification of individuals based on physical characteristics has recently gained popularity and falls under the category of pattern recognition. Biometric recognition has emerged as an effective strategy for preventing security breaches, as no two people share the same physical characteristics. "Gait recognition" specifically refers to identifying individuals based on their walking patterns. Human gait is a method of locomotion that relies on the coordination of the brain, nerves, and muscles. Traditionally, human gait analysis was performed subjectively through visual observations. However, with advancements in technology and deep learning, human gait analysis can now be conducted empirically and without the need for subject cooperation, enhancing the quality of life. Deep learning methods have demonstrated excellent performance in human gait recognition. In this article, the authors employed the VGG19 transfer learning model for human gait recognition. They used the public benchmark dataset CASIA-A for their experimental work, which contains a total of 19,139 images captured from 20 individuals. The dataset was segmented into two different patterns: 70:30 and 80:20. To optimize the performance of the proposed model, the authors considered three hyperparameters: loss, validation loss (val_loss), and accuracy rate. They reported accuracy rates of 96.9% and 97.8%, with losses of 2.71% and 2.01% for the two patterns, respectively.

Many researchers have preferred non-invasive techniques for recognizing the exact type of physiological abnormality in the vocal tract by training machine learning algorithms with feature descriptors extracted from the voice signal. However, until now, most techniques have been limited to classifying whether a voice is normal or abnormal. It is crucial that the trained Artificial Intelligence (AI) be able to identify the exact pathology associated with voice for implementation in a realistic environment. Another issue is the need to suppress the ambient noise that could be mixed up with the spectra of the voice. Current work proposes a robust, less time-consuming and non-invasive technique for the identification of pathology associated with a laryngeal voice signal. More specifically, a two-stage signal filtering approach that encompasses a score-based geometric approach and a glottal inverse filtering method is applied to the input voice signal. The aim here is to estimate the noise spectra, to regenerate a clean signal and finally to deliver a completely fundamental glottal flow-derived signal. For the next stage, clean glottal derivative signals are used in the formation of a novel fused-scalogram which is currently referred to as the "Combinatorial Transformative Scalogram (CTS)." The CTS is a time-frequency domain plot which is a combination of two time-frequency scalograms. There is a thorough investigation of the performance of the two individual scalograms as well as that of the CTS database.Nine classification metrics are used to investigate performance, which are: sensitivity, mean accuracy, error, precision, false positive rate, specificity, Cohen’s kappa, Matthews Correlation Coefficient, and F1 score. Implementation of the VOice ICar fEDerico II (VOICED) standard database provided the highest mean accuracy of 94.12(%) with a sensitivity of 93.85(%) and a specificity of 97.96(%) against other existing techniques. The current method performed well despite the data imbalance that exists between classes.

Over the past few years, blockchain technology has gained significant attention. This surge in popularity can be attributed to the emergence of cryptocurrencies and the development of smart contracts. Cryptocurrency is a digital currency that eliminates the problem of double spending. Cryptocurrencies like Bitcoin, Ethereum, Litecoin, Stellar, Zcash, Maker, Aave, etc. become popular and are preferred for money transfers. Smart contracts are the next popular technology on the blockchain after cryptocurrency. It can be considered a piece of code that can execute automatically when the predefined conditions are fulfilled. Researchers believe that the potential of blockchain with smart contracts is only in its initial stages and that its true potential has yet to be fully discovered. Hence, an extensive bibliometric analysis is conducted to understand blockchain trends for smart contracts and to give future directions in this field. For this analysis, various steps are followed, starting with formulating the research question, defining the scope of our research, extracting and analyzing data, answering the research question, and finally, drawing a conclusion. This research paper will be fruitful for scholars and researchers, providing an extensive statistical and network analysis of extracted smart contracts publications.

Object grasping is an important skill for robots to interact with the real world, especially in unstructured environments where occlusions and different shapes of target objects are present. In this work, we introduce a robot grasping pipeline called SUGrasping, which can obtain the grasping poses more precisely for target objects. The grasping pipeline treats the Truncated Signed Distance Function (TSDF) and point clouds of the grasping scene as input simultaneously. The proposed multi-head 3D U-Net accepts reconstructed TSDF representation and outputs the grasping configurations, including predicted grasp quality, orientation and width of the gripper. The point cloud is fed into PointNet to obtain the semantic segmentation results for all objects in the grasping workspace. With the help of point cloud inside the gripper, the relationship between the gripper and semantic information can be established. It makes robots know which object they are grasping, rather than just removing objects in the workspace like previous works. Experimental results show that the proposed method has an improvement in grasping success rate and percent cleared of target objects, which outperforms state-of-the-art methods compared in this paper.

In this study, we integrate the Bidirectional Encoder Representations from Transformers (BERT) model with the Cycle Generative Adversarial Network (CycleGAN) to create a system for Chinese text style transfer. Natural language processing (NLP) involves converting human languages into data interpretable by computers, enabling applications like text classification, chatbots, and dialogue systems. Recent advancements, such as Google's transformer model and the BERT technique, have significantly improved NLP capabilities through self-attention mechanisms and unsupervised pretraining. Text style transfer modifies the style of texts without altering their semantics. Previous methods like StyIns and models based on disentangled representation learning highlight the challenges of retaining text meaning during style transfer. Our system leverages CycleGAN’s unsupervised learning to convert unpaired data between wuxia and fantasy styles while preserving semantics. Using the pretrained BERT model from the Chinese Knowledge and Information Processing (CKIP) Lab, our experimental results demonstrate successful style conversion, maintaining the original meanings of texts. This integration of BERT and CycleGAN shows promise for further advancements in NLP applications.

As dialects are widely used in many countries, there is growing interest in incorporating them into various applications, including conversational systems. Processing spoken dialects is an important module in such systems, yet it remains a challenging task due to the lack of resources and the inherent ambiguity and complexity of dialects. This paper presents a comparison of two approaches for identifying spoken Maghrebi dialects, tested on an in-house corpus composed of four dialects: Algerian Arabic Dialect (AAD), Algerian Berber Dialect (ABD), Moroccan Arabic Dialect (MAD), and Moroccan Berber Dialect (MBD), as well as two variants of Modern Standard Arabic (MSA): MSA_ALG and MSA_MAR. The first method uses a fully connected neural network (NN2) to retrain several Transfer Learning (TL) models with varying layer numbers, including Residual Networks (ResNet50, ResNet101), Visual Geometric Group networks (VGG16, VGG19), Dense Convolutional Networks (DenseNet121, DenseNet169), and Efficient Convolutional Neural Networks for Mobile Vision Applications (MobileNet, MobileNetV2). These models were chosen based on their proven ability to capture different levels of feature abstraction: deeper models like ResNet and DenseNet are capable of capturing more complex and nuanced patterns, which is critical for distinguishing subtle differences in dialects, while VGG and MobileNet models offer computational efficiency, making them suitable for applications with limited resources. The second approach employs a “stacked generalization” strategy, which merges predictions from the previously trained models to enhance the final classification performance. Our results show that this cascade strategy improves the overall performance of the Language/Dialect Identification system, with an accuracy increase of up to 5% for specific dialect pairs. Notably, the best performance was achieved with DenseNet and ResNet models, reaching an accuracy of 99.11% for distinguishing between Algerian Berber Dialect and Moroccan Berber Dialect. These findings indicate that despite the limited size of the employed dataset, the cascade strategy and the selection of robust TL models significantly enhance the system’s performance in dialect identification. By leveraging the unique strengths of each model, our approach demonstrates a robust and efficient solution to the challenge of spoken dialect processing.

Unsupervised Domain Adaptation (UDA) in person re-identification (reID) addresses the challenge of adapting models trained on labeled source domains to unlabeled target domains, which is crucial for real-world applications. A significant problem in clustering-based UDA methods is the noise in pseudo-labels generated due to inter-domain disparities, which can degrade the performance of reID models. To address this issue, we propose the Unsupervised Dual-Teacher Knowledge Distillation (UDKD), an efficient learning scheme designed to enhance robustness against noisy pseudo-labels in UDA for person reID. The proposed UDKD method combines the outputs of two source-trained classifiers (teachers) to train a third classifier (student) using a modified soft-triplet loss-based metric learning approach. Additionally, a weighted averaging technique is employed to rectify the noise in the predicted labels generated from the teacher networks. Experimental results demonstrate that the proposed UDKD significantly improves performance in terms of mean Average Precision (mAP) and Cumulative Match Characteristic curve (Rank 1, 5, and 10). Specifically, UDKD achieves an mAP of 84.57 and 73.32, and Rank 1 scores of 94.34 and 88.26 for Duke to Market and Market to Duke scenarios, respectively. These results surpass the state-of-the-art performance, underscoring the efficacy of UDKD in advancing UDA techniques for person reID and highlighting its potential to enhance performance and robustness in real-world applications.